Sustained Inattentional Blindness: the Role of Location in the Detection of Unexpected Dynamic Events
Total Page:16
File Type:pdf, Size:1020Kb
PSYCHE an interdisciplinary journal of research on consciousness Sustained Inattentional Blindness: The Role of Location in the Detection of Unexpected Dynamic Events Steven B. Most, Daniel J. Simons, Brian J. Scholl, & Christopher F. Chabris Department of Psychology Harvard University 33 Kirkland Street Cambridge, MA 02138 U.S.A. [email protected] [email protected] Copyright (c) Steven B. Most, Daniel J. Simons, Brian J. Scholl, & Christopher F. Chabris 2000 PSYCHE, 6(14), December 2000 http://psyche.cs.monash.edu.au/v6/psyche-6-14-most.html KEYWORDS: Inattentional blindness, perception, attention, selective looking, attentional capture, vision, visual cognition. ABSTRACT: Attempts to understand visual attention have produced models based on location, in which attention selects particular regions of space, and models based on other visual attributes (e.g., in which attention selects discrete objects or specific features). Previous studies of inattentional blindness have contributed to our understanding of attention by suggesting that the detection of an unexpected object depends on the distance of that object from the spatial focus of attention. When the distance of a briefly flashed object from both fixation and the focus of attention is systematically varied, detection appears to have a location-based component. However, the likelihood that people will detect an unexpected event in sustained and dynamic displays may depend on more than just spatial location. We investigated the influence of spatial location on inattentional blindness under precisely controlled, sustained and dynamic conditions. We found that although location-based models cannot fully account for the detection of unexpected objects, spatial location does play a role even when displays are visible for an extended period. 1. Introduction One central issue in the study of visual attention involves the question of whether attention is solely location-based or whether it can also be affected by non-spatial factors. Each view encompasses several variants. For example, location-based models of attention include the "spotlight" model (Posner, 1980; see Cave & Bichot, 1999, for a review), in which attention acts to "illuminate" whatever falls within an attended region, and the "zoom lens" model (Eriksen & St. James, 1986), in which attention can be directed either at broad areas in coarse detail or at small areas in fine detail. Models based on other visual attributes include those suggesting that discrete objects or features within a scene can be directly attended, unmediated by a spatial spotlight (e.g., Baylis & Driver, 1993; Duncan, 1984; see Scholl, in press, for a review). Recent studies of inattentional blindness have explored whether the detection of an unexpected object can inform us about the nature of attention (Mack & Rock, 1998; Newby & Rock, 1999). Specifically, does detection depend on the spatial proximity to the focus of attention? In this paper, we also explore the role of proximity to the focus of attention in the detection of unexpected objects, but unlike previous studies, we use a dynamic, sustained inattentional blindness task (Most et al, in press). This task is well-suited to exploring on-going perceptual events under controlled conditions. Location-based models might predict that the probability of noticing an unexpected object will be affected by the distance of the object from the focus of attention (Newby & Rock, 1999). If an unexpected object appears near the focus of attention, then observers will be much more likely to see it than if it appears far away from the focus of attention. However, detection should not be affected by the similarity of the unexpected object to the other objects in the display. Evidence for an object-based component to selection comes from findings that multiple features are more readily processed when they are part of a single object than when they are part of multiple objects, even when the features are the same distance apart in both cases (e.g., Duncan, 1984; Watson & Kramer, 1999). Feature-based models argue that attention is preattentively directed to simple features in the display (e.g., red or bright) in addition to objects or regions. For example, if observers were attending to the red objects in a display, a new red object would be noticed but a new black object might not be. Accordingly, feature-based models might predict that unexpected objects will be detected provided that they are featurally similar to objects already being attended, regardless of their position in the display (Most et al, in press). This view also predicts that people may not necessarily see unexpected objects that have different properties from already-attended objects, even if the unexpected object appears in close proximity to the focus of attention. Recently, the "inattentional blindness" paradigm has contributed to our understanding of the effects of spatial proximity on the detection of unexpected objects (Mack & Rock, 1998; Newby & Rock, 1998). "Inattentional blindness" (IB) refers to the finding that observers who are engaged in attentionally demanding tasks often fail to see unexpected objects or events. In a typical task, observers viewed a cross that appeared in a computer display for 200 milliseconds before it was replaced by a patterned mask. They were asked simply to judge whether the horizontal or the vertical arm of the cross was longer. On a "critical trial," an additional unexpected object appeared simultaneously with the cross, usually in one of the quadrants defined by the cross's arms. The dependent measure was whether or not observers retrospectively reported having consciously seen the unexpected object. Depending on the particular variant of this task, between 25% and 75% of the observers failed to notice the unexpected object (Mack & Rock, 1998). Mack and Rock (1998), inspired partly by "spotlight" metaphors of attention (Posner, 1980), hypothesized that unexpected events occurring anywhere within a contiguous "zone of attention"-where the zone's diameter was roughly defined by the length of the cross's arms-would more likely be seen than unexpected events occurring outside this zone. If this hypothesis is correct, then objects appearing within one of the quadrants of the cross should be seen just as frequently as objects appearing on one of the cross's arms. Their results showed that 80% of observers saw the unexpected object (put another way, the level of inattentional blindness, IB, was 20%) both when the unexpected object appeared on an arm of the cross and when it appeared in one of the cross's quadrants. In further experiments, observers were far less likely to detect an unexpected object that appeared outside the region defined by the horizontal and vertical extent of the cross (or of a rectangle that was sometimes substituted for the cross) than to detect an object that appeared within that region. Taken together, these findings seem to support a location-based model of attention over an object-based model, as well as supporting the idea of a contiguous "zone of attention" (Mack & Rock, 1998). However, these findings do not eliminate the possibility that increasing levels of IB were due to the increasing distance from the center of attention, rather than to the qualitative difference between appearing within or outside the "zone of attention." Extending this work, Newby and Rock (1998) directly examined the effect of distance on detection rate by parametrically increasing the distance of the unexpected object from the center of the cross. Furthermore, by having the attended cross appear away from fixation, they were able to vary the distance of the unexpected object from the center of the cross while keeping its distance from fixation constant. Under these conditions, distance from the center of attention, and not from fixation, accounted for the resulting rate of noticing: when the object appeared further away from the focus of attention (i.e., the cross), fewer observers saw it than when it appeared at the center of focal attention. However, their evidence did not support the notion of a bordered "zone of attention." Instead, the decrease in detection rate with distance occurred in a continuous fashion, with no sudden drop-off beyond the region defined by the cross. Although the evidence from these IB studies suggests a role for spatial proximity in mediating the likelihood of detecting an unexpected object, these effects might not generalize to the perception of realistic, dynamic scenes; the displays used in these studies typically involved briefly presented and masked static stimuli. It is noteworthy, then, that studies involving more sustained and dynamic displays may support models in which attention can be directed at discrete objects, features, or event sequences in addition to spatial locations. For example, extending other work on "selective looking" (Becklen & Cervone, 1983), Simons and Chabris (1999) conducted a series of experiments in which observers watched a video of two teams of basketball players-one team clad in white shirts, the other in black shirts-each passing a basketball among themselves. The observers were instructed to count the number of passes made by either the white team or the black team. Partway through this task, either a woman with an umbrella or a person in a gorilla costume unexpectedly walked through the center of the action, remaining clearly visible for about five seconds before exiting the display. The observers were then asked if they had seen the unexpected object. Thirty-five percent of the observers failed to notice the woman with the umbrella, even though her presence was obvious to anyone not engaged in the counting task. Perhaps more startling, given its more unusual nature, even more people failed to notice the gorilla (56%). In both cases the unexpected figure moved through the same spatial locations that were being occupied by the attended basketball players. Thus, an appeal to solely spatially based models of attention cannot explain the high degree of IB.